In this post, Digoal presents his predictions for what we'll see for PostgreSQL in 2020.
Before we get into what exactly I think will be the three biggest things to look out for in 2020 from PostgreSQL. Let's have a little recap of what happened over the past year.
In 2019, at Alibaba Cloud we have witnessed the rapid buildup and deployment of Alibaba Cloud's ApsaraDB RDS for PostgreSQL, our enterprise-grade database hosting service for PostgreSQL. Here's some things that happened.
We were the first to release a commercial version of PostgreSQL 12 and provide global coverage. Compared with earlier versions, PostgreSQL 12 provides significant improvements in terms of online transaction processing, analysis performance, and general functionality. Already, several enterprises across various industries have chosen to make PostgreSQL 12 part of their cloud systems.
We released a super large instance type, which supports up to 32 terabytes in storage, 1 million IOPS, and 4 Gbit/s read and write speeds.
We released the PostgreSQL SGX encrypted version. This version prevents data leakage in the network, operating system, and database, as well as in storage, data center, and on cloud platform to protect the security of each and every user's data. Compared with transparent data encryption (TDE), link encryption and cloud disk encryption provide a higher level of security, meaning it can meet the high-level security demands of the financial sector.
We launched the zero data loss feature for data protection. With this feature, you can select synchronous, semi-synchronous, or asynchronous data protection based on your business needs.
- Also, we also released the real-time precision marketing (crowd labeling) plug-in roarmingbitmap, the high-dimensional vector plug-in pase for image recognition and search, the chemical analysis plug-in rdkit, the database scheduled task plug-in pg_cron, the log_fdw plug-in that supports real-time log query for the current instance through SQL, the MySQL external table plug-in mysql_fdw, and the PostgreSQL external table plug-in postgre_fdw.
We enhanced the database ecosystem. Alibaba Cloud Data Transmission Service (DTS) supports all versions starting from 9.4, as well as the incremental migration of PostgreSQL and the incremental synchronization of MySQL to RDS PostgreSQL.
However, as the saying goes, today's best performance is tomorrow's bare minimum. Even the current epidemic cannot stop the pace progress. So what major features will 2020 offer to PostgreSQL users around the world?
Prediction 1: More Major Customers Will Move to the Cloud
My first set of predictions relates to the trend that more and more large enterprises, which happen also to be some of PostgreSQL's major customers, will be moving to the cloud this year. Large enterprises usually have their own technical systems, and therefore their requirements for cloud databases are quite different from those of small and medium-sized enterprises (SMEs). Given these facts, Alibaba Cloud as well as PostgreSQL need to take certain steps to facilitate the migration of major customers to the cloud. Therefore, my predictions are as follows:
Better Security Compliance
- The PostgreSQL SGX encrypted version avoids mutual trust issues between customers and cloud vendors and ensures user data security by comprehensively preventing data leakage in networks, operating systems, databases, storage, data centers, and cloud platforms. SGX Phase I is difficult to use and requires the modification of applications. So, in 2020, the SDK experience will be optimized, reducing the application modification costs.
- Standard SSD encryption was introduced in 2019. In 2020, full coverage for enhanced SSD storage will be added as well as support for CMK.
- In 2020, a disaster tolerance and backup feature and quasi-real-time secondary instances will be provided to meet the cross-regional data center needs of large enterprises.
Better and Mutual Integration
In 2020, open and independent clusters for PostgreSQL's major customers will be supported. When major customers migrate to the cloud, one key task is to integrate the customer's technology stack with the cloud services of the cloud vendor. In 2020, the exclusive cluster will be launched for PostgreSQL's major customers. This is a user-controlled, highly customizable, and cost-effective independent RDS cluster, which offers the fully managed format of RDS while providing open integration capabilities:
Custom instance distribution rules: For example, instances can be distributed to different compute nodes based actual business circumstances.
Custom overselling ratio: For example, for testing or non-core business, you can use a high overselling ratio. Meanwhile, you can use a no-oversell mode for core business. This can allow you to improve the overall cost performance.
- Users can install custom plug-ins.
- Database super admin account and operating system permissions will be opened so that the customer's database admins (DBAs) can exercise more comprehensive control over the instances. For example, no tickets are required in the case of urgent matters.
- The customer's own management platform will be able to be integrated, which allows the customer to continue using the capabilities of their original technology stacks.
Exclusive clusters for major PostgreSQL customers can meet the database customization needs of large enterprises.
An Improved Sharding Function
- To handle businesses that have systems that require frequent reads and writes and that have a large amount of data, users generally need to partition them in the business layer, which is a difficult process. The sharding functions of PostgreSQL, such as citus and pg-xc, also demand a high level of technical skills. However, a sophisticated, stable, and easy-to-use PostgreSQL sharding product is on the horizon. In 2020, the PostgreSQL sharding function will be released, one that can provide PostgreSQL customers with better instance scalability and meet the needs of businesses with larger data volumes and higher concurrency.
- In 2020, read-only instances will be made easier to use and allow primary instances of any type to create read-only instances. This will meet the needs of users who require read scaling.
Large enterprises need to deploy their business across clouds for disaster tolerance between different clouds. Cross-cloud database disaster tolerance will be the first step toward cross-cloud management.
Prediction 2: The Application Microservice Architecture Will Enhance Database Multitenancy
With the development of microservice architectures, businesses can be divided into smaller modules using PostgreSQL. Each microservice may correspond to one database. This gradually will decrease the size of individual databases but exponentially increase the number of instances. Such an architecture will have a huge impact on an enterprise's database management and O&M costs.
- With database multitenancy, one instance will manage thousands of databases. This solution will meet performance and stability requirements while also supporting resource isolation and resource quotas.
- With database multitenancy, batch instance management capabilities will be improved.
This will be especially applicable to enterprises that provide SaaS services, such as platform-based service, ERP, and CRM.
Prediction 3: More Services Will Move Online in 2020
In 2019, we set up a DingTalk group for the PostgreSQL technology exchange, which has more than 3,600 members now. Over 40 technical live broadcasts were shared throughout the year, and many technical experts from the Alibaba and PostgreSQL communities answered online questions. From time to time, PostgreSQL technology sharing experts are selected. We expect this group will surpass 10,000 members in 2020.
So, what other surprises might 2020 have in store for us? Well, let's wait and see!